[Trac #286] Wireless oops (usb_hcd_submit_urb list_add corruption during firmware load)

[Zarro Boogs per Child]

#286: Wireless oops (usb_hcd_submit_urb list_add corruption during firmware load)
----------------------+-----------------------------------------------------
 Reporter:  dwmw2     |        Owner:  marcelo 
     Type:  defect    |       Status:  reopened
 Priority:  high      |    Milestone:  BTest-3 
Component:  wireless  |   Resolution:          
 Keywords:            |  
----------------------+-----------------------------------------------------
Changes (by marcelo):

  * status:  closed => reopened
  * resolution:  worksforme =>

Comment:

 Actually I've hit it the other day. Its _very_ rare though.

 This specific sequence

 [  631.398419]  [<c89f85ef>] wlan_add_card+0x0/0x204 [usb8xxx]
 [  631.430125]  [<c89f8258>] wlan_remove_card+0x0/0x15f [usb8xxx]
 [  631.462425]  [<c8a0402e>] libertas_sbi_register+0x2b/0x35 [usb8xxx]
 [  631.495866]  [<c89f80e3>] wlan_init_module+0x29/0x3a [usb8xxx]
 [  631.528970]  [<c042a1e8>] sys_init_module+0x1474/0x15cb

 Indicates its if_usb_disconnect->wlan_remove_card (if_usb_disconnect is
 not on the backtrace, though, why?). Now for the USB disconnect method to
 be called you need either a call to usb_unregister() or usb_reset_device()
 (the former is being done by the probe function).

 My guess is that the probe and disconnect functions are executing
 simultaneously.

 However I don't know how that can happen.

 Reopening for reference.

-- 
Ticket URL: <http://dev.laptop.org/ticket/286#comment:8>
One Laptop Per Child <http://laptop.org/>